Fix compatibility issues

fedebotu commented 3 years ago

Here is a changelog of the fixes:

Change LQRStep from autograd.Functional to nn.Module for compatibility (more info about legacy autograd function with non-static method here and here)
Commented out save_for_backward (not needed in nn.Module apparently since it is using the Functional API for saving the weights by default, more info here)
LQRStep returns backward and forward trajectory by default
Add local import in test files
Add Circle CI testing (you can change the account info later)
Miscellaneous fixes

bamos commented 3 years ago

Thanks for the updates! Are the gradient tests passing with this? I think replacing the Function with the Module will make it so the right backward pass doesn't get called (here) and will make the derivatives silently wrong

bamos commented 2 years ago

Fixed in https://github.com/locuslab/mpc.pytorch/commit/bf19cd4405d8242e93f4c0a5464739f98798a833

locuslab / mpc.pytorch

Fix compatibility issues #32